home · mobile · calendar · colloquia · 2010-2011 · 

Colloquium - Clauset

The Mathematics of Global Terrorism
Department of Computer Science

Terrorism is often treated by the media and academics as completely senseless, highly strategic, and largely contingent (that is, unpredictable). Thus, it is commonly assumed that no fundamental laws exist and that forecasting is generally impossible. For some aspects of terrorism, however, these assumptions turn out to be false. In particular, for the frequency and severity of deadly terrorist attacks, there exist strongly regular patterns, which may admit simple mechanistic explanations.

In this talk, I'll summarize my recent work on characterizing and explaining two robust patterns extracted from large data sets of terrorist attacks: (i) a power-law relationship that describes the frequency of severe attacks, in which the probability of an attack that kills x individuals decays like Pr(x) ∝ x-α with α = 2.4 ± 0.1, and (ii) a universal acceleration in attack frequency by terrorist organizations, in which the typical delay τ between the kth and (k+1)th deadly attack by a group decays like τ = k-β with β = 1.10 ± 0.02.

I'll conclude by arguing that these patterns and statistical algorithms in general can help us make principled and potentially accurate long-term forecasts of trends in and risks from global terrorism.

Aaron Clauset is an Assistant Professor of Computer Science at the University of Colorado Boulder and a Fellow in the Colorado Initiative for Molecular Biotechnology. He holds a BS in Physics from Haverford College, a PhD in Computer Science from the University of New Mexico, and was an Omidyar Fellow at the Santa Fe Institute. His research is broadly focused on developing the fields of computational social science and computational systems biology. His work uses tools from computer science, physics and statistics to analyze large data sets, characterize their structure and build computational models to explain their origin and forecast their evolution. His results have appeared in Nature, Science, JACM, STOC, AAAI, SIAM Review, Physical Review Letters, and the Journal of Conflict Resolution.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)